def get_coverage(self, x, y, width, height):
assert(not self.obj.ignored)
return image.get_coverage(
self.obj.surface.surface,
self.matrix,
x, y, width, height
)
def get_coverage(self, x, y, width, height):
assert(not self.obj.ignored)
return image.get_coverage(
self.obj.surface.surface,
self.matrix,
x, y, width, height
)
def recognize(self):
class Quadrilateral():
"""This class iterates through small box areas in a quadriliteral.
This is usefull because some scanners have trapezoidal distortions."""
# Assumes top left, top right, bottom right, bottom left
# corner.
def __init__(self, p0, p1, p2, p3):
self.x0 = p0[0]
self.y0 = p0[1]
self.x1 = p1[0]
self.y1 = p1[1]
self.x2 = p2[0]
self.y2 = p2[1]
self.x3 = p3[0]
self.y3 = p3[1]
# 0 -> 1
self.m0 = (self.y1 - self.y0) / (self.x1 - self.x0)
self.m1 = (self.x2 - self.x1) / (self.y2 - self.y1)
self.m2 = (self.y3 - self.y2) / (self.x3 - self.x2)
self.m3 = (self.x0 - self.x3) / (self.y0 - self.y3)
self.top = min(self.y0, self.y1)
self.bottom = max(self.y2, self.y3)
self.left = min(self.x0, self.x3)
self.right = max(self.x1, self.x2)
def iterate_bb(self, step_x, step_y, test_width, test_height,
padding):
y = self.top
while y + test_height < self.bottom:
x = self.left
while x + test_width < self.right:
yield x, y
x += step_x
y += step_y
def iterate_outline(self, step_x, step_y, test_width, test_height,
padding):
# Top
x, y = self.x0, self.y0
x += padding
y += padding
dest_x = self.x1 - padding - test_width
dest_y = self.y1 + padding
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x**2 + dist_y**2)
for step in xrange(int(length / step_x)):
yield x + dist_x * step / (
length / step_x), y + dist_y * step / (length / step_x)
yield dest_x, dest_y
# Bottom
x, y = self.x3, self.y3
x = x + padding
y = y - padding - test_height
dest_x = self.x2 - padding - test_width
dest_y = self.y2 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x**2 + dist_y**2)
for step in xrange(int(length / step_x)):
yield x + dist_x * step / (
length / step_x), y + dist_y * step / (length / step_x)
yield dest_x, dest_y
# Left
x, y = self.x0, self.y0
x += padding
y += padding
dest_x = self.x3 + padding
dest_y = self.y3 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x**2 + dist_y**2)
for step in xrange(int(length / step_y)):
yield x + dist_x * step / (
length / step_y), y + dist_y * step / (length / step_y)
yield dest_x, dest_y
# Right
x, y = self.x1, self.y1
x = x - padding - test_width
y = y + padding
dest_x = self.x2 - padding - test_width
dest_y = self.y2 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x**2 + dist_y**2)
for step in xrange(int(length / step_y)):
yield x + dist_x * step / (
length / step_y), y + dist_y * step / (length / step_y)
yield dest_x, dest_y
def iterate(self, step_x, step_y, test_width, test_height,
padding):
for x, y in self.iterate_bb(step_x, step_y, test_width,
test_height, padding):
ly = self.y0 + self.m0 * (x - self.x0)
if not ly + padding < y:
continue
ly = self.y2 + self.m2 * (x - self.x2)
if not ly - padding > y + test_height:
continue
lx = self.x1 + self.m1 * (y - self.y1)
if not lx - padding > x + test_width:
continue
lx = self.x3 + self.m3 * (y - self.y3)
if not lx + padding < x:
continue
yield x, y
for x, y in self.iterate_outline(step_x, step_y, test_width,
test_height, padding):
yield x, y
bbox = None
img = self.obj.sheet.get_page_image(self.obj.page_number)
if img is None or img.recognize.matrix is None:
self.obj.sheet.valid = 0
return
x = self.obj.x
y = self.obj.y
width = self.obj.width
height = self.obj.height
# Scanning area and stepping
step_x = defs.textbox_scan_step_x
step_y = defs.textbox_scan_step_x
test_width = defs.textbox_scan_width
test_height = defs.textbox_scan_height
# extra_padding is always added to the box side at the end.
extra_padding = defs.textbox_extra_padding
scan_padding = defs.textbox_scan_uncorrected_padding
quad = Quadrilateral((x, y), (x + width, y), (x + width, y + height),
(x, y + height))
try:
quad = Quadrilateral(
*img.recognize.find_box_corners(x, y, width, height))
# Lower padding, as we found the corners and are therefore more acurate
scan_padding = defs.textbox_scan_padding
except AssertionError:
pass
surface = img.surface.surface
matrix = img.recognize.matrix
for x, y in quad.iterate(step_x, step_y, test_width, test_height,
scan_padding):
# Use the image module directly as we are calling in *a lot*
coverage = image.get_coverage(surface, matrix, x, y, test_width,
test_height)
if coverage > defs.textbox_scan_coverage:
if not bbox:
bbox = [x, y, test_width, test_height]
else:
bbox_x = min(bbox[0], x)
bbox_y = min(bbox[1], y)
bbox[2] = max(bbox[0] + bbox[2], x + test_width) - bbox_x
bbox[3] = max(bbox[1] + bbox[3], y + test_height) - bbox_y
bbox[0] = bbox_x
bbox[1] = bbox_y
if bbox and (bbox[2] > defs.textbox_minimum_writing_width
or bbox[3] > defs.textbox_minimum_writing_height):
# Do not accept very small bounding boxes.
self.obj.data.state = True
self.obj.data.x = bbox[0] - (scan_padding + extra_padding)
self.obj.data.y = bbox[1] - (scan_padding + extra_padding)
self.obj.data.width = bbox[2] + 2 * (scan_padding + extra_padding)
self.obj.data.height = bbox[3] + 2 * (scan_padding + extra_padding)
else:
self.obj.data.state = False
self.obj.data.x = self.obj.x
self.obj.data.y = self.obj.y
self.obj.data.width = self.obj.width
self.obj.data.height = self.obj.height
def recognize(self):
class Quadrilateral():
"""This class iterates through small box areas in a quadriliteral.
This is usefull because some scanners have trapezoidal distortions."""
# Assumes top left, top right, bottom right, bottom left
# corner.
def __init__(self, p0, p1, p2, p3):
self.x0 = p0[0]
self.y0 = p0[1]
self.x1 = p1[0]
self.y1 = p1[1]
self.x2 = p2[0]
self.y2 = p2[1]
self.x3 = p3[0]
self.y3 = p3[1]
# 0 -> 1
self.m0 = (self.y1 - self.y0) / (self.x1 - self.x0)
self.m1 = (self.x2 - self.x1) / (self.y2 - self.y1)
self.m2 = (self.y3 - self.y2) / (self.x3 - self.x2)
self.m3 = (self.x0 - self.x3) / (self.y0 - self.y3)
self.top = min(self.y0, self.y1)
self.bottom = max(self.y2, self.y3)
self.left = min(self.x0, self.x3)
self.right = max(self.x1, self.x2)
def iterate_bb(self, step_x, step_y, test_width, test_height, padding):
y = self.top
while y + test_height < self.bottom:
x = self.left
while x + test_width < self.right:
yield x, y
x += step_x
y += step_y
def iterate_outline(self, step_x, step_y, test_width, test_height, padding):
# Top
x, y = self.x0, self.y0
x += padding
y += padding
dest_x = self.x1 - padding - test_width
dest_y = self.y1 + padding
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x ** 2 + dist_y ** 2)
for step in xrange(int(length / step_x)):
yield x + dist_x * step / (length / step_x), y + dist_y * step / (length / step_x)
yield dest_x, dest_y
# Bottom
x, y = self.x3, self.y3
x = x + padding
y = y - padding - test_height
dest_x = self.x2 - padding - test_width
dest_y = self.y2 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x ** 2 + dist_y ** 2)
for step in xrange(int(length / step_x)):
yield x + dist_x * step / (length / step_x), y + dist_y * step / (length / step_x)
yield dest_x, dest_y
# Left
x, y = self.x0, self.y0
x += padding
y += padding
dest_x = self.x3 + padding
dest_y = self.y3 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x ** 2 + dist_y ** 2)
for step in xrange(int(length / step_y)):
yield x + dist_x * step / (length / step_y), y + dist_y * step / (length / step_y)
yield dest_x, dest_y
# Right
x, y = self.x1, self.y1
x = x - padding - test_width
y = y + padding
dest_x = self.x2 - padding - test_width
dest_y = self.y2 - padding - test_height
dist_x = dest_x - x
dist_y = dest_y - y
length = math.sqrt(dist_x ** 2 + dist_y ** 2)
for step in xrange(int(length / step_y)):
yield x + dist_x * step / (length / step_y), y + dist_y * step / (length / step_y)
yield dest_x, dest_y
def iterate(self, step_x, step_y, test_width, test_height, padding):
for x, y in self.iterate_bb(step_x, step_y, test_width, test_height, padding):
ly = self.y0 + self.m0 * (x - self.x0)
if not ly + padding < y:
continue
ly = self.y2 + self.m2 * (x - self.x2)
if not ly - padding > y + test_height:
continue
lx = self.x1 + self.m1 * (y - self.y1)
if not lx - padding > x + test_width:
continue
lx = self.x3 + self.m3 * (y - self.y3)
if not lx + padding < x:
continue
yield x, y
for x, y in self.iterate_outline(step_x, step_y, test_width, test_height, padding):
yield x, y
bbox = None
img = self.obj.sheet.get_page_image(self.obj.page_number)
if img is None or img.recognize.matrix is None:
self.obj.sheet.valid = 0
return
x = self.obj.x
y = self.obj.y
width = self.obj.width
height = self.obj.height
# Scanning area and stepping
step_x = defs.textbox_scan_step_x
step_y = defs.textbox_scan_step_x
test_width = defs.textbox_scan_width
test_height = defs.textbox_scan_height
# extra_padding is always added to the box side at the end.
extra_padding = defs.textbox_extra_padding
scan_padding = defs.textbox_scan_uncorrected_padding
quad = Quadrilateral((x, y), (x + width, y), (x + width, y + height), (x, y + height))
try:
quad = Quadrilateral(*img.recognize.find_box_corners(x, y, width, height))
# Lower padding, as we found the corners and are therefore more acurate
scan_padding = defs.textbox_scan_padding
except AssertionError:
pass
surface = img.surface.surface
matrix = img.recognize.matrix
for x, y in quad.iterate(step_x, step_y, test_width, test_height, scan_padding):
# Use the image module directly as we are calling in *a lot*
coverage = image.get_coverage(surface, matrix, x, y, test_width, test_height)
if coverage > defs.textbox_scan_coverage:
if not bbox:
bbox = [x, y, test_width, test_height]
else:
bbox_x = min(bbox[0], x)
bbox_y = min(bbox[1], y)
bbox[2] = max(bbox[0] + bbox[2], x + test_width) - bbox_x
bbox[3] = max(bbox[1] + bbox[3], y + test_height) - bbox_y
bbox[0] = bbox_x
bbox[1] = bbox_y
if bbox and (bbox[2] > defs.textbox_minimum_writing_width or
bbox[3] > defs.textbox_minimum_writing_height):
# Do not accept very small bounding boxes.
self.obj.data.state = True
self.obj.data.x = bbox[0] - (scan_padding + extra_padding)
self.obj.data.y = bbox[1] - (scan_padding + extra_padding)
self.obj.data.width = bbox[2] + 2 * (scan_padding + extra_padding)
self.obj.data.height = bbox[3] + 2 * (scan_padding + extra_padding)
else:
self.obj.data.state = False
self.obj.data.x = self.obj.x
self.obj.data.y = self.obj.y
self.obj.data.width = self.obj.width
self.obj.data.height = self.obj.height
def get_coverage(self, x, y, width, height):
return image.get_coverage(
self.obj.surface.surface,
self.matrix,
x, y, width, height
)